**1. Introduction**

McEwen and Akil [1] recently outlined major steps of progress from 50 years of research on the neurobiology of stress. The concept of stress responses as adaptive mechanisms capable of being overworked or overloaded, thereby producing an "allostatic load" that increases propensity for disease, is one of many contributions from the McEwen laboratory [1,2]. Repeated exposure to even mild stressors can constrain adaptive mechanisms, contributing to cardiometabolic, cognitive, and behavioral disorders [1–3]. Such approaches to modeling physiological dysregulation predict not only illness but also longevity [4,5]. Moreover, psychosocial stress is associated with elevated oxidative stress [6,7], a further contributor to disease and aging. The high prevalence of stress-related diseases naturally motivates the search for effective interventions.

Accumulated research on the standardized Transcendental Meditation® (TM®) technique suggests it can reduce unwanted effects of stress. This technique was revived from the ancient Vedic tradition and made available to the world at large by Maharishi Mahesh Yogi starting in the 1950s [8,9]. The first research on physiological effects appeared in 1970 [10]. As reported and reviewed by others, investigations of the TM program as a therapeutic intervention have reported benefits for PTSD [11–13], anxiety disorders [14], and risk factors for cardiovascular disease (CVD) [15–18]. Moreover, studies on healthcare utilization sugges<sup>t</sup> reductions in a wide spectrum of diseases [19,20]. Evidence supports the conclusion that the automatic self-transcending nature of this technique bears primary responsibility for these effects [21,22].

Research on long-term molecular and cellular effects of stress that relate to disease and aging has focused on the immune system and mitochondrial energy production. Early investigations found differences affecting glucocorticoid and inflammatory signaling [23]. Further research led to the discovery of a "Conserved Transcriptional Response to Adversity (CTRA)" accompanying severe or chronic stress [24,25]. The CTRA involves upregulation of pro-inflammatory genes and downregulation of antiviral and antibody components of the defense response [26].

Energy metabolism also is a key component of stress responses and adaptation [27], and mitochondrial energy production is affected by stress [28,29]. Both chronic stress and aging are associated with reductions in mitochondrial function and energy efficiency [30,31].

Some mind–body interventions have been reported to decrease or reverse such effects of stress [32–34], including effects on the "epigenetic clock", a reproducible biomarker of biological aging [35]. However, meditation studies and other mind–body intervention research, especially studies examining transcriptomic effects, have tended to be of short duration [36,37]. This leaves a gap in our knowledge of effects deriving from decadeslong practice of these programs. Based on prior research indicating that effects of the TM program are cumulative, we hypothesized that long-term practice would produce transcriptomic differences connected to the program's stress-reducing and health-promoting benefits. Results of this study appear to support that hypothesis. Due to budgetary constraints and to this being the first of its kind, the study is exploratory, not definitive, in nature. Nevertheless, it employs the three steps characterizing gene expression comparisons, namely a discovery step (microarray comparison [38]), a verification and validation step applying a quantitatively more accurate approach (qPCR) to a sample of genes, and a functional analysis showing consistency with prior research on stress effects and meditation. The results appear to provide promising avenues for further investigation.

#### **2. Materials and Methods**

#### *2.1. Research Design and Participants*

This study used DNA microarray transcriptomics and qPCR technologies in peripheral blood mononuclear cells (PBMCs) to compare non-practitioner control groups and TM practitioner groups. First, two demographically well-matched small groups were selected from a larger pool of volunteers for the microarray analysis (Table 1). Later, qPCR analysis of 15 genes selected from those differentially expressed in the microarray was used to test the reproducibility of the microarray results in the larger pool of 45 volunteers. All study participants were recruited through advertising on the campus of Maharishi International University and in public places in or near Fairfield, Iowa. The design and methods were approved by the University's Institutional Review Board. Following both written and oral descriptions of the study, participants gave signed consent prior to participation.

**Table 1.** Demographic Matching of Control and Transcendental Meditation® (TM) Groups for Microarray Analysis.


\* Subjective socioeconomic status (SES), using a 5-point scale from 1 for "lower class" to 5 for "upper class".

To reduce genetic variation, study participants were limited to self-identified white males and females. Prospective participants were excluded if they reported a doctoridentified history of diabetes, nerve damage, heart attack, coronary heart disease, stroke, kidney failure, cancer, any other life-threatening illness, a major psychiatric disorder, or substance abuse. In addition, candidates for the control group were excluded if they had ever been instructed in the TM program. Practitioners of the TM program were excluded if they had not regularly practiced the program twice a day or usually twice a day.

The meditation program consisted of the standard TM technique [9] practiced in the microarray TM group for 458 ± 49 months, with later addition of the TM-Sidhi® program, also practiced twice daily in this group, for 406 ± 50 months. The TM-Sidhi program, like the TM program, is drawn from the ancient Vedic tradition and is said to promote more rapid incorporation of the benefits of TM practice into daily life [9]. As with the small groups for microarray analysis, the larger groups used for qPCR analysis were not significantly different in age (Table 2). The TM group was notably higher in the number of vegetarians and tended to be higher in other indicators of healthy lifestyle. Subjective socioeconomic status (SES) data were not available. TM in this qPCR study was practiced for 475 ± 40 months, and the TM-Sidhi program for 396 ± 44 months.


**Table 2.** Demographic Matching of Control and TM Groups for qPCR Validation Analysis.

\* Subjective SES data were not available for these participants.

#### *2.2. PBMC Preparation*

Blood was drawn in random order from 4–6 participants a day between 10 AM and 4 PM by a certified phlebotomist using a 19-gauge butterfly needle. A total of 16 mL was drawn in two collection tubes (BD Vacutainer® CPTTM Mononuclear Cell Preparation Tube). The buffy coat containing the PBMCs was harvested according to the manufacturer's instructions, and the cell pellet was stored at −80 ◦C.

#### *2.3. RNA Extraction, Concentration Measurement, and Integrity Check*

RNA was extracted from the buffy coat employing the RNAzol B Kit (Ambion®). RNA concentration was estimated using a UV absorption ratio method [39]. Ratios above 1.7 were considered sufficiently pure for the sample to be used for microarray and qPCR analyses. RNA integrity was analyzed by automated electrophoresis on microfluidic labchips using the ExperionTM System (Bio-Rad). RNA Quality Indicator (RQI) values considered acceptable were in the range 7 < RQI ≤ 10. Samples meeting the criterion for integrity were selected for array-based expression profiling.

#### *2.4. Whole-Genome mRNA Expression Using Bead-Based Array*

Total RNA samples from matched TM and control groups were shipped on dry ice to the Genomics Core at the University of Chicago, Chicago, IL, for genome-wide mRNA expression profiling using Illumina® Gene Expression BeadChip technology. A minimum amount of 50 ng of total RNA for each sample was labeled using the Illumina TotalPrepTM RNA Labeling Kit (Ambion). Labeled samples were incubated on Illumina HumanHT-12v4 Expression BeadChips containing probes for 47,231 features and imaged using the Illumina iScan system. Raw data from the iScan were converted using the GenomeStudio software package and Gene Expression Module (Illumina), resulting in identification of 16,247 genes and loci.

Microarray data were subjected to statistical analysis using R BioConductor (lumi and BeadArray) packages that include methods correcting for Type 1 and Type 2 errors [40]. Differential expression for each individual gene was considered acceptable if it met comparatively strict criteria, i.e., ratio of TM and control group gene-expression values (normalized by the quantile method) ≥2.0 and *p*-value for differential expression ≤0.05. A heat map with hierarchical clustering was created from data on the differentially expressed genes using the average linkage method and the Pearson distance metric to show how data aggregated. This analysis was performed in R version 2.11.1 with the gplots package using hclust and heapmap.2 functions [41]. Networks with molecular paths plausibly affected by practice of the TM program were identified using Ingenuity® Pathway Analysis (IPA®) software [42].

To assign functional annotations for differentially expressed genes, gene ontological enrichment analysis was performed through the DAVID database, a free online web-based tool. Differentially expressed genes submitted to DAVID were sorted into lists of ranked genes under enriched gene ontological process terms. The *p*-values displayed are a DAVID adjustment of the Fisher exact test *p*-value from testing for a significantly higher number of genes in the submitted list belonging to the group of genes, compared with all genes in the human genome [43].

A similar approach was used in the disease association analysis, in this case using Genomatix. Genomatix is data-mining software for extracting and analyzing gene relationships from literature databases such as NCBI PubMed and annotation data such as Gene Ontology. The software calculates overrepresentation of specific biological terms within the input and ranks genes related to specific diseases in the output. The program calculates *p*-values in a similar manner to that described for the DAVID online tool.

#### *2.5. qPCR Analysis*

Target genes chosen for validation of the microarray results included genes representing the main findings of the microarray component, including key relationships to health and aging. The primers were designed using Primer BLAST and were purchased from IDT® Technologies (http://www.idtdna.com/site, accessed 5 January 2014). The target genes, the reference gene, and their primer sequences are shown in Table S1: Primer Sequences for qPCR Reactions.

First, for each participant, complementary DNA (cDNA) was constructed from the extracted RNA employing the iScriptTM cDNA Synthesis Kit (Bio-Rad). Each cDNA reaction mixture of 20 μL contained 1 μg of the RNA template and 10 μL of master mix (containing 1.0 μL iScript reverse transcriptase solution, 4.0 μL of 5× iScript mix, and 5 μL nuclease-free water). The PCR reaction was run in the GeneAmp® PCR 9700 systen (Applied Biosystems) in the following steps: 5 min at 25 ◦C; 30 min at 42 ◦C; 3.5 min at 85 ◦C, and hold at 4 ◦C.

Second, the qPCR experiment was conducted using the SsoFastTM EvaGreen® Supermix (Bio-Rad). Each reaction contained 5 μL 1× SsoFast EvaGreen supermix, a final concentration of 320 nM each of forward primer and reverse primer, 5 ng cDNA template, and 2.9 μL nuclease-free water in a 10-μL final volume. Finally, samples were assayed in quadruplicate in a 320-well reaction plate using a C1000 thermal cycler (Bio-Rad) combined with a CFX 384 Real-Time System (Bio-Rad) in the following cycling steps: 1. Enzyme activation: 95 ◦C, 30 s, 1 cycle; 2. Denaturation: 95 ◦C, 3 s, 40 cycles; 3. Annealing/Extension: 58 ◦C, 5 s, 40 cycles; 4. Melt curve: 65–95 ◦C (in 0.5 ◦C increments, 5 s/step), 1 cycle.

The relative expression level of each gene ("fold difference") was calculated using the (2.0 (ΔΔCt)) method [44]. Statistical analysis was performed using ANOVA in SPSS. An alpha level of *p* ≤ 0.05 was adopted for statistical significance.
